Historically web filters have been used to filter pornography on the Internet by blocking websites based on URL blacklists and lexical analysis. These techniques are becoming less effective as the Internet is becoming more dynamic. They also are ineffective at addressing pornographic videos which can be found on media sharing sites that would be considered legitimate by a blacklist filter. Lexical analysis is also ineffective as online videos often have little text associated with them that would indicated the videos contents is pornographic.
One solution to this may be to intercept the video stream and analyse the content of the video using computer vision techniques. However video analysis is time consuming and delaying the video while it is being analysed can be detrimental to the user experience and cause connection problems.
Therefore, there is a desire for an improved video analysis system which is less time consuming.
It is an object of the present invention to provide a video analysis system which overcomes the disadvantages of the prior art, or at least provides a useful alternative.